NOUMENOpen app
NOUMEN

moondream · Balanced

Moondream 3 Preview

Added Oct 16, 2026

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Moondream 3 Preview is the original 9B mixture-of-experts vision-language checkpoint with 2B active parameters, for visual question answering, captioning, OCR, and visual reasoning. It accepts text or one image with optional reasoning, a 32,000-token context, and up to 4,096 output tokens.

  • Reasoning
  • Vision
Input
Text, Images
Output
Text
Discovery tier
Balanced

Pricing

USD per 1M tokens
Input
$0.30
Output
$2.50

Reference prices as published by the model provider. API example

Specifications

Context window
32K
Max output
4.1K
Average output (7 days)
94 tokens
Longer than 9% of 468 models

Side by side

Moondream 3 Preview compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Moondream 3 PreviewmoondreamThis modelQwen3 VL 235B A22B Instruct OriginalQwenRemove
ProvidermoondreamQwen
Discovery tierBalancedBalanced
AddedOct 16, 2026Sep 25, 2025
Limits
Context window32K—
Max output4.1K32.77KBest
Average output94 tokens288 tokens
Pricing, USD per 1M tokens
Input$0.30$0.30
Output$2.50$1.90Best
Performance
Request success (24h)—100.0%
Benchmark indexes
Intelligence Index—9.9
Math Index—70.7
Capabilities
ReasoningYesNo
VisionYesYes
Reasoning effort——

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Moondream 3 Preview
  • Qwen3 VL 235B A22B Instruct Original

Intelligence Index

  • Moondream 3 Preview—
  • Qwen3 VL 235B A22B Instruct Original9.9
3.8Catalog models by score55.1

Indexes

  • Math Index

    Composite of the math evaluations

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 70.7

Document reasoning

  • AA-LCRv1.1

    Reasoning across long documents

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 32.7%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 6.6%

  • GPQA Diamond

    Graduate-level science questions

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 71.2%

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 82.3%

  • IFBench

    Following precise output instructions

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 42.7%

Coding

  • LiveCodeBench

    Recent competitive programming problems

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 59.4%

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 70.7%

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 6.8%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 35.1%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Moondream 3 Preview: —
    • Qwen3 VL 235B A22B Instruct Original: 32.7%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "moondream/moondream3-preview",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="moondream/moondream3-preview",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

moondream

Other moondream models, starting with the same family.